1. Field of the Invention
The invention relates to the field of computer systems, and, more particularly, to data compression in such.
2. Description of Background
Data compression in networks which transmit data between computer systems often involves transmitting data using an encoding scheme to reduce the amount of bandwidth and receiver buffer space necessary when compared to transmitting the same data not using the encoding scheme. Lossless data compression is typically an encoding scheme which can regenerate all of the original data. The Lempel-Ziv Stac (“LZS”) algorithm is an example of a lossless data compression encoding scheme.
The LZS algorithm is a standard approach for achieving lossless data compression. This algorithm searches an incoming data stream for repeating data patterns by comparing the incoming data intended for transmission to a history buffer, or dictionary, maintained at the transmitter side of the link. Repetitive data patterns are replaced by a token of shorter length; this token usually consists of a pointer to the repetitive data's matching location in the history buffer and the length of the matching data sequence.
For example, the transmitter might maintain a history of the past 1 Kbyte of data frames, and whenever a new data sequence is input to the compressor, this history is searched for matching data sequences. Instead of transmitting the actual data, a token might be generated which effectively says “go back 150 bytes in the history and transmit the next 10 bytes”. At the receive side of the link, a decompressor recreates an exact replica of the compressor's dictionary and uses the tokens provided by the data stream to invert the compression operation.
Thus the tokens can be viewed as pointers into the 1 Kbyte data history. The history files at both ends of the link should match, otherwise the decompressor pointer token will output meaningless data. Thus, future data is compressed based on the previous data history.
Typical compression ratios vary with the type of data and amount of redundancy present. Completely random data will not be compressed very much at all, while highly redundant traffic with a lot of idle characters may experience a very large improvement.
Data compression efficiency thus depends on the characteristics of the network communications protocol which is employed. In particular, this invention is concerned with an inter-system channel (ISC) link which is used to transmit data functions within a Parallel Sysplex computer architecture.